Cell-based clustering using pixel clusters: Gaussian blur sigma=2, k=30, bin=0.8

Heatmaps of pixel clusters x markers

Compare pixel clusering with cell clustering

Cell-based FlowSOM clustering using all pixel clusters

No transformation

Normalize by total (covert from count to frequency), arcsinh transform, 99.99% normalization

Drop pixel clusters that do not have at least one marker ranked in top 5, normalize by total, arcsinh, 99.99% normalization

## [1] "Number of dropped clusters: 52"

## [1] "Dropped clusters: 1,2,3,4,5,6,7,8,9,10,11,12,13,14,16,17,19,20,23,25,26,32,34,37,39,44,45,48,49,55,56,59,61,64,65,67,68,71,73,77,78,79,80,82,83,84,85,86,87,88,89,90"

Cell-based FlowSOM clustering using pixel hierarchical clusters

No transformation

Normalize by total (covert from count to frequency), arcsinh transform, 99.99% normalization

Drop pixel clusters that do not have at least one marker ranked in top 2, normalize by total, arcsinh, 99.99% normalization

## [1] "Number of dropped clusters: 9"

## [1] "Dropped clusters: 5,6,9,10,11,12,22,25,27"